News:

"The phone is a remarkably complex, simple device,
and very rarely ever needs repairs, once you fix them." - Dan/Panther

Main Menu

Dial testing and measuring, what’s available / how to do. Links to other threads

Started by dsk, November 29, 2018, 12:47:57 PM

Previous topic - Next topic

dsk

I hope this thread may be a guide to find the easiest way to measure your dials according to your needs.  Not the most accurate, if you do not need it. Maybe the one who might be most easy to use with your equipment, I also hope others will help with their ways to do it, and please make a thread for each way of testing a dial, and just make a short description, and link to that thread here.

Short descriptions as dial a 0 on another well working dial at a known well working dial, and release both at the same time to compare the time to full return. May not need a thread, or maybe...


What do you actually need? How accurate is good enough? (for you)

The make /break ratio are usually OK and does seldom need adjustments, the nominal speed are usually 10 pulses pr second  +/- 7% but most exchanges accepts much larger deviations than that, maybe from 8 to 18.
my accepts at least 8-12, but some SIP lines needs it almost exact.

The less accurate and easiest way is to just measure the time used for returning the dial from 0 to stop, and 1.2- 1.5 seconds use to be OK

So to the tools The BK Precision 1045 or 1050 has a go or not go limit witch probably are based on what is OK by US standards around 1970.
http://www.classicrotaryphones.com/forum/index.php?topic=16046.msg166113#msg166113
The most accurate I have used so far is actually the setup using audacity and a spreadsheet, at least after I have found and corrected some errors in the spreadsheet. 
http://www.classicrotaryphones.com/forum/index.php?topic=17033.msg175673#msg175673

I have also got my selves a Siemens dial tester, it prints out a paper tape with the dial results graphical, and then we have to decode and calculate, or just compare with a picture of the ideal printout. This is excact, and has been done by Siemens since before 1930.
http://www.classicrotaryphones.com/forum/index.php?topic=21462.msg218785#msg218785

The last I have tested is a small free German software, it is not as accurate as the audacity, but probably good enough in almost every case.
You need a PC, the software and 2 wires on a serial port. It has a great display, it is so easy to read out the data.
http://www.classicrotaryphones.com/forum/index.php?topic=21454.0



dsk

kleenax

Didn't know if I really should "reply" to this thread, but here it is....

This is what I use to do my phones & dials. Can't say that I have ever seen this "NORTHEAST ELECTRONICS CORP" dial tester Model TTS-26B mentioned, but I may have missed it somewhere here.

REF:  http://www.classicrotaryphones.com/forum/index.php?topic=12231.msg130447#msg130447
Ray Kotke
Recumbent Casting, LLC

dsk


dsk

The ease of using the different testers may be pretty different, and so do the accuracy.

We are easily accepting a reading from a meter metering something witch are to complicated to determine if it may be right or not.
I have now tried to measure a push-button pulse telephone with the tree units mentioned above, does the difference in readings be caused by the inaccuracy of the dial, or the metering?  How accurate does it need to be?

By my experience dial speeds from 9-12 pps has been accepted by all my equipment, and break percentage from less than 50 to 80 has worked.  What does not go well?  The BTMC dial 7006A!When normal dials had equal length of the breakes, this had 5 times longer last break, and for my equipment that caused a on-hook signal or a hook flash signal.
dsk

dsk

Done a lot of experimenting lately, and even when it is not a scientific answer, I am pretty sure that the readings vary deending on other components in the circuit.  What is the most stable solution seems to be audacity and spreadsheet. The best results seems to be when I add on a tone of at least 800 Hz. 1000Hz works fine  :) and the readings are pretty accurate.  Without that tone it is slightly harder to identify the exact time of pulse start and end.

If I disconnect the dial and just measures that all the others worked well too.  ;)

dsk

TelePlay

I have never used an extraneous tone in the testing circuit when testing dial speed using Audacity. I have a 25' line cord from my bench to my computer so whatever is on my bench can be tested and I have never had a problem determining the first 9 break/make spikes using a long line cord.

I am posting this because Audacity has several "volume" functions that allow one to increase the height of the spikes when capturing a pulsing dial. What I have noticed is the dial itself has something to do with spike heights. Might be size of pulse contacts or the physical pressure exerted on the contacts by their springs. Don't know. I use my MIC in jack to record the pulse spikes.

Below are images for a WE 7C (500), 6C (302) and Trimeline (probably a #9) dial. The Trimline was the weakest but the Audacity processing functions lets one improve the spikes so the contact break/make times are easily identified. Of the 3 dials tested, the Trimline dial produced the smallest spikes.

The last image is a comparison showing 9 break/make cycles for each dial (the 10th can never be found in that the make tails off with opening of the muting contacts. So, 9 cycles are used and normalized to 10 (9 cycles in 0.9 seconds is the same as 10 cycles in 1.0 seconds).

The processing functions include amplify (and audacity automatically calculates the amount of amplification needed to take the biggest spike to 100%, be it a upward or downward spike).

Instead of using amplify, one can use normalize which can be set for any dB normalization (-3.0 or 0.0) work best - trial and error).

Using normalize alone also increased background noise. Audacity has a noise reduction function where a background sample is take (3 or 4 seconds) and that is then used to reduce the noise in the recording. The noise reduced recording can then be normalized (or amplified) to make small spikes large, easier to see.

Seems using the Amplify function alone is all that is needed to make pulse spikes bigger.

The last image is a comparison of all 3 dial pulse recording using only the Audacity amplify function. Wasn't really needed for the 7C and 6C dial but it did improve the Trimline dial spike heights. The 9 pulse cycles for each dial are numbered with the first break being 0 and the last break, the beginning of the 10th cycle, being 9. There is a 10th make spike but no 10th break spike due to opening of the mute contacts.

If I took the phone or the dial to my computer desk and used 18" clip leads to hook the dial up to the computer, the spikes would be a bit higher so there really is not much loss over 25' of line cord.

Finally, dials can be tested stand alone out of the phone by attaching the clip leads to the dial pulsing terminals, to the L1 and L2 network terminals if the dial is in the phone but the line cord is missing or if the phone has not been opened by attaching the test clip leads directly to the red/green line cord leads (be they a 4 prong plug or a modular plug stuck into a jack with exposed terminals). Testing a dial while in the phone, as received, is an immediate indication if the dial is fine or will need cleaning.

I doubt few test dials and those that do it professionally use a Sage. This is added to dsk's topic in case anyone interested in dial analysis and cleaning cares to use Audacity, which as dsk said, is the best way to do it if you don't have a Sage.

dsk

Great pictures and explanations. The tone is not a must, but it helps me to skip all that filtering job. (That maybe not is to much of job when you know how to do it. )

At least audaciity is still giving the best picture of how your dial is working.

dsk

TelePlay

Amplitude processing once the pulses are recorded is real easy and fast to do, nothing complicated.

Start by highlighting the part of the recording you want to amplify,

click "Effect" on the top menu,

click "Amplify" in the drop down box.

( A box will open showing how many dB's the selection will be amplified to do a maximum amplification without clipping,
     the software calculates the amplification factor needed for maximum amplification. )


Simply click "OK" and that's it.

markosjal

I read recently (on an official AE Document) that some AE dials were designed to run at 12PPS


Good idea on the tone for testing pulse duration.
Phat Phantom's phreaking phone phettish

TelePlay

Quote from: markosjal on January 02, 2021, 05:31:10 AM
Good idea on the tone for testing pulse duration.

While it's really up to the person doing the dial analysis and adjustment, a tone is not necessary for rotary dials. It just takes additional equipment and set up and one runs the risk of damage to the computer audio input card. Much easier to connect the two dial pulse terminals to the lead and ground wires of a micro plug inserted into the mic input of a PC or laptop with Audacity installed.

Was working on a 6A dial last week that was running below 9.5 PPS so took it out for cleaning and lubrication. After that, it ran a bit over 11 PPS. Adjusted the governor and in a rare moment, after 3 or 4 adjustments, ended up with exactly 10.0 PPS. Since the governor adjustment is trial and error, hitting 10.0 exactly is a very rare event.

The image below is the cropped raw Audacity wave form showing the time scale at the top. The image was annotated to show the 9 pulse cycles used to determine the dial speed and also shows the Break and Make points recorded as audio spikes when the dial pulse leaves open and close.

There is no 10th contact "Break" in that after the 10th contact "Make", the dial opens and closes the talk circuit leaves, the other terminals on the back of a rotary dial to get the phone into its talk mode. There are always 9 full Break/Make cycles so the time it takes to completer those cycles is divided by 9 to get the dial speed normalized to 10 pulses.

In this case, it took 0.90 seconds for the dial to complete 9 pulse cycles. That speed is normalized to 10 pulse cycles by dividing the number of complete pulse cycles, 9, by 0.09 it took to complete the 9 pulse cycles and that yields 10 Pulses Per Second. Dialing 0 or 10 pulses at 10 PPS takes exactly 1.0 second. If it takes more than 1 second, that indicates a slow dial. If it takes less than a 1 second, that means a fast dial.

One can use an analog clock with a second hand to quickly check a dial. Simply dial 0 and wait for the second had to move. As soon as it does, release the finger wheel and watch to see if the dial returns to zero before (fast dial) or after (slow dial) the next second ticks off. Wait a few seconds before releasing the dial to the "test" second tick can be anticipated and the dial released right on the tick. With a little practice, it is easy to do as a fast and dirty assessment of a dial.

Rarely if ever are fast dials found as purchased or received. Dials slow down for 2 reasons:  1) the lubrication has dried up and friction slows the dial, and 2) the dial at one point was slowing down and over lubricated which resulted in the excess oil collecting dust and dirt over time which hardened the excess lubrication in gear teeth physically slowing down, and in some cases stopping, on its return to 0.

Fast dials are usually found after slow dial disassembly, cleaning and properly lubricating the dial during assembly. This indicates a situation in which the dial was slowing down during it original use period and a field technician, a telephone repair person, adjusted the governor in the field to speed up the pulses per second to get the dial within specifications. Properly cleaning and lubricating one of these "field adjusted slow dials" will almost always result in a fast dial, usually over 11 PPS.